Anal. Chem. 1995,67, 3509-3514
Direct Analysis of Enzymatic Reactions of Oligosaccharides in Human Serum Using Matrix=AssistedLaser Desorption Ionization Mass Spectrometry Randy M. Whiial, Monica M. Palcic, Ole Hindsgaul, and Llang Li* Department of Chemistty, University of Alberta, Edmonton, Alberta, Canada T6G 2G2
Matrix-assisted laser desorption ionization mass spectrometry has been developed for direct mass analysis of enzymatic reaction products of oligosaccharidesin human blood serum without the use of extraction or chromatographic separation. Molecular labeling of the Substrate is used to achieve both the detection sensitivity and selectivity required in rapid analysis of reactionproducts in serum. It is found that tetramethylrhodamine (TMR)labeled oligosaccharidesprovide 100-foldsensitivity enhancement over the corresponding underivatized oligosaccharides. In order to selectively retain the "R-labeled molecules on the sample probe while washing away contaminants in a serum sample, a sample/matrix preparation method is developed. This technique provides detection sensitivity of labeled oligosaccharides in the range of hundreds of femtomoles per microliter. The mass measurement accuracy is better than 0.01%when a linear time-of-flightmass spectrometer is used. The application of the technique is illustrated for the subpicomole detection and quantitation of the conversion of the disaccharide aFuc(142)/3Gal-"MFtto the blood group B active trisaccharide a.Fuc(142)[aGal( 1-3)lpGal-m catalyzed by the blood group B galactosyltransferase present in human serum. Mass spectrometry using fast atom bombardment (FAB), electrospray, and matrix-assisted laser desorption ionization O I ) has recently been developed for the analysis of products of enzymatic reactions in biological fluids. For example, FAB mass spectrometry has been reported for analyzing endogenous metenkephalin and P-endorphm from their respective precursors in human cerebrospinal fluid' and dynorphm A, dynorphm B, and a-neodynorphin from the prodynorphin precursor in human pituitary.2 More recently, MALDI time-of-flight mass spectrometry has been shown to be an effective method for mass analysis of the products of in vitro processing involving dynorphin A3 and cyclosporin A4 in human blood and neuropeptide Y in human cerebrospinal f l ~ i d Electrospray .~ mass spectrometry has been used as an assay for peptidyl-a-hydroxyglycinea-amidating ~~~
(1) Liu, D.; Wood, G. W.; Desiderio, D. M.J. Chromatogr. 1990,530, 235252. (2) Silberring, J.; Brostedt, P.; Thomwall, M.; Nyberg, F.J Chromatogr, 1991, 554, 83-90. (3) Chou, J. 2.; Kreek, M. J.; Chait, B. T.J Am. Sot. Mass Spectrom. 1994,5, 10-16. (4) Muddiman, D. C.; Gusev, A I.; Proctor, A; Hercules, D. M.; Venkataramanan, R; Diven, W. Anal. Chem. 1994,66, 2362-2368.
0003-2700/95/0367-3509$9.00/0 0 1995 American Chemical Society
lyase enzyme? Although these techniques require an extensive sample cleanup (except that of Costello et al.5) through extraction, filtration, or chromatographic separation, the mass spectrometric approach provides high molecular spec5city for the detection and idenacation of the reaction products. We are exploring the use of the MALDI technique in the area of glycobiology. One of the major observations driving research in this new field is that the structures of the carbohydrate chains present on cell-surface glycoproteins and glycolipids present sites hr the binding of viruses and bacteria, in addition to their roles in both normal and abnormal development and in cell-cell adhesion. In particular, developing embryonic cells, activated cells, and tumor cells d produce altered carbohydrate sequences on their cell surfaces, whose functions are only beginning to be under~tood.~ These altered carbohydrate structures appear to be the product of abnormal expression of the class of enzymes termed glycosyltransferases, the enzymes that control1 the assembly of oligosaccharides.* The ability to monitor changes in the activity of these enzymes in cells and tissues is of prime importance in understanding the regulation of their expression, and ideally, the required assays of enzyme activities should be simple and applicable to crude cell or tissue homogenates, including blood and serum. The assay of glycosyltransferases in cells or tissues usually involves quantitation of the transfer of 3H-or I3C-labeled sugars from their sugar nucleotides to either an isolated or a synthetic oligosaccharide acceptor.gJ0 While the technique is sensitive (down to -1 pmol), there is no product characterization. This means that one cannot decide if one or more sugars have been added to the acceptor or if it is a degradation product of the acceptor that in fact becomes a substrate for another interfering glycosyltransferase. To partially overcome these difficulties, radioactive products can be separated by high-performance liquid chromatography (HPLC) and characterized by coelution with authentic standards, if these are available.11J2There are distinct (5) Costello, C.E.; Juhasz, P.; Ekman, R; Heilig, M.; Agren, H. Proceedings of the 42nd ASMS Conferenceon Mass Spectromety and Allied Topics, Chicago, Illinois, May 29-June 3, 1994; p 25. (6) Unsworth, E. J.; Treston, A M. Proceedings ofthe 42nd ASMS Conference on M a s Spectromety and Allied Topics, Chicago, Illinois, May 29-June 3, 1994; p 154. (7) Varki, A Glycobiology 1993,3, 97-130. (8) Schachter, H. In Molecular Glycobiology; Fukuda, M., Hindsgaul, O., Eds.; IRL Press: Oxford, U q 1994; pp 88-162. (9) Sadler, J. E.; Beyer, T. A; Oppenheimer. C. L;Paulson, J. C.; Prieels, J. P.; . Methods Enzymol. 1982,83, 458-514. Rearick, J. I.; Hill, R L (10) Palcic, M. M.; Heerze, L. D.; Pierce, M.; Hindsgaul, 0. Glycoconjugate J. 1988,5,49-63.
Analytical Chemistty, Vol. 67, No. 19, October 1, 1995 3509
advantages to such chromatographic methods, but they remain very time consuming. Replacement of the radiolabels with fluorescent probes can result in dramatic increases in sensitivity, but again, either HPLC or electrophoresis is needed to gain evidence for formation of the expected p r ~ d u c t s . ~Finally, ~-~~ immunological-basedmethods, where product formation is both detected and structurally characterized by a specific protein, are very powerful and sensitive (100 fmol). Such methods, however, require the production of either a refined polyclonal antiserum or a monoclonal antibody or the availability of a sequencespecific le~tin.'~-~~ Reported herein is the development of a MALDI method for the rapid and sensitive detection of the product of a glycosyltransferase enzyme reaction. Blood group B galactosyltransferase enzyme occurs naturally in the serum of blood group B individuals. This enzyme transfers a galactose residue from uridine diphosphogalactose (UDP-Gal) in a-linkage to OH-3 of the galactose residue in oligosaccharide chains terminating in the sequence aFuc(l-2)pGal producing the blood groupB antigenic determinant aFuc(1-2) [aGal(l-3)lpGal. This is an ideal model system for the development of sensitive assays for the products of enzymes involved in oligosaccharide biosynthesis or degradation since this enzyme is present in very low abundance in an extremely crude sample: human serum. Serum of individuals lacking this enzyme activity, e.g., blood group 0 or A individual!!, is additionally available as a control. The fast MALDI assay presented in this report represents one of the so-called mass tracer methods we are currently developing. By analogy to well-established radioassay tracer and fluorescent tracer techniques, mass tracer methods are based on mass spectrometry for monitoring chemical, enzymatic, or other changes in biological systems, with no or minimal sample preparation in detection. In this MALDI method, an appropriate functional group is linked to the target molecules, and then the MALDI technique is used to follow the transformation of the labeled molecular species. After experimenting with different labeling groups, we find that tetramethylrhodamine WR) derivatives of oligosaccharides are -1Wfold more sensitive than underivatized oligosaccharides. Moreover, we have developed a sample preparation method for mixing samples and matrix for on-probe sample cleanup. With this method, salts, buffers, and other potential interfering contaminants in serum are washed away, and yet the TMR-labeled species are selectively retained on the sample probe. (11) Koenderman, A H. L;Wijermans, P. W.; Van den Eijnden, D. H. FEBS Lett. 1987,222, 42-46. (12) Brockhausen, I.; Carver, J. P.; Schachter, H. Biochem. Cell. Biol. 1988,66, 1134-1151. (13) Honda, S.; Iwase, S.; Makino, A; Fujwara, S. Anal. Biochem. 1989,176, 72-77. (14) Lee, K. B.; Desai, U. R; Palcic, M. M.; Hindsgaul, 0.;Linhardt, R J. Anal. Biochem. 1992,205, 108-114. (15) Zhao, J. Y.; Dovichi, N. J.; Hindsgaul, 0.; Gosselin S.; Palcic, M. M. Glycobiology 1994,4, 239-242. (16) Zhang, Y.; Le, X;Dovichi, N. J.; Compston, C. A; Palcic, M. M.; Diedrich, P.; Hindsgaul, 0. Anal. Biochem., in press. (17) Stults, C. L.; Wilbur, B. J.; Macher, B. A. Anal. Biochem. 1988,174, 151156. (18) Palcic, M. M.; Ratcliffe, R W.; Lamontagne, L. R; Good, A H.; Alton, G.; Hindsgaul, 0. Carbohydr. Res. 1990,196, 133-140. (19) Crawley, S. C.; Hindsgaul, 0.; Alton, G.; Pierce, M.; Palcic, M. M. Anal. Biochem. 1990,185, 112-117. (20) Zatta, P. F.; Nyame IC; Cannier, M. J.; Mattox, S. A; Prieto, P. A; Smith, D. F.; Cummings, R D. Anal. Biochem. 1991,194, 185-191. (21) Spohr, U.; Morishima, N.; Hindsgaul, 0.;Lemieux, R U. Can. J. Chem. 1985, 63,2664-2668.
3510 Analytical Chemistry, Vol. 67,No. 19, October 1, 1995
Chart 1 Structure aFuc(l-2)pGal
Abbreviation
-
aFuc( 1+2)[aGal( 1+3)]pGal
disac-
-
0-
-
aFuc(l~2)PGal(l+3)[aFuc(l-+4)]PGlcNAc
Leb-
MATERIALS AND METHODS
Preparation ofhbeled Oligosaccharides. Chart 1lists the structures and their respective abbreviations for the labeled oligosaccharides that pertain to this work. Leb-OCH321and LebMC0,2I as well as aFuc(l--Z),!?Gal-MCO (disac-MCO) and aFuc(1-2) [aGal(l-3) ]/?Gal-MCO @-MCO),22were available from earlier work. The conversion of MCO glycosides to their T M R derivatives was performed as previously described.'5J6 Enzyme Reactions. F i e nanomoles of aFuc(l-2)~Gal-TMR (acceptor) in aqueous solution at 1 mg mL-I was added to an Eppendorf tube and lyophilized to dryness. After drying, 50 pL of human serum (blood group B), 2 pL of 2.5 mM UDP-Gal (donor), 0.6 p L of 0.5 M MnC12, and 5 pL of 0.5 M sodium cacodylate buffer (PH 7.1) were added. The mixture was incubated at 37 "C for 90 min, and an aliquot was used to check for the presence of product through MALDI analysis. The introduction of the TMR substituent has a minor effect on the galactosyltransferase reaction. Under the conditions employed for the in vitro conversion, the rate of reaction of the labeled species was found to be 90%that of the parent aFuc(l-2)j3Gal0 (CH&CH3 by standard radiochemical assay methods.1° MALDI Sample Preparation. All nonserum samples were prepared for MALDI analysis using the dried droplet method of sample preparation (see, for example, Hillenkamp et alF3 >. 2 , s Dihydroxybenzoic acid (2,SDHB) matrix solution was prepared at a concentration of 10 mg mL-' in 3096 acetonitrile/water. a-Cyano-4hydroxycinnamicacid (4-HCCA) matrix solution was prepared as a saturated solution (-7 mg mL-') in 30%acetonitrile/ water. The matrix solutions were mixed on a vortex mixer for 1 min and then centrifuged to remove undissolved matrix crystals. A 0.5 p L aliquot of matrix solution and 0.5-1.0 p L of analyte solution were placed on the sample probe and allowed to dry. For samples analyzed from buffered human serum, CHCCA was used as the matrix. CHCCA matrix solution was prepared at a concentration of 20 mg mL-' in 70% acetonitrile/water. A 0.5 pL aliquot of matrix solution was applied to the probe tip and allowed to dry. The highconcentration matrix solution produced a dense layer of matrix on the sample probe of -2 mm diameter, giving a surface density of 17 nmol/mm2. The serum sample was ~
~~~~~
(22) Lemiew, R U. Chem. Soc. Reo. 1978,7, 423-452. (23) Hillenkamp, F.; Karas, M.; Beavis, R C.; Chait, B. T. Anal. Chem. 1991, 63. 1193A-1203A
diluted in half with 50% ethanol/water. On top of the matrix solution, 0.5-1.0 pL of the serum supernatant was applied. Just before the sample dried (-1 min), the probe tip was dipped into pure room temperature water for 45 s. Excess water was removed by gently touching a wiper to the edge of the sample probe. Washing the probe tip just before the sample dries proved more effective at removing the high concentration of buffers and other species present in serum. The washed probe tip was then inserted into the mass spectrometer. TOFMS. A linear, time-lag focusing TOF mass spectrometer was used to collect all mass spectra. The MALDI instrument was designed and constructed at the University of Alberta. A detailed description of this system will be reported elsewhere. In brief, it consists of an acceleration region containing a repeller plate and an ion extraction plate to which 12 kV voltages are applied. After the laser desorption/ionization takes place, a pulsed voltage (1 kv) is applied to the repeller for ion extraction. The time delay between the ionization and ion extraction events is optimized at 280 ns and used throughout the work presented herein. The total flight tube length is about 1m. A nitrogen laser (Photochemical Research Associates LNlOOO) is used to generate the MALDI ions. The mass spectrum is captured using either a LeCroy 9310M or 9350M digital oscilloscope. The major difference is the sampling rate. The 9350M scope provides a sampling rate of up to 1 x lo9 samples/s or a time resolution of 1ns/point, whereas the 9310M scope has a time resolution of 10 ns/point. High-resolutionmass spectra are collected on the 9350M oscilloscope. All massspectra are the result of signal averaging of 100 laser shots. Signal averaging was performed on the oscilloscope without selection. The averaged mass spectrum was then downloaded to a PC for processing.
L ~ ~ - M C+O ~ a ' 889.0
- 1
600
800
1000
1200
1400
MI2 Flgure I. MALDI mass spectrum of a mixture of three derivatives of an oligosaccharide using 2,5-DHB as the matrix: 10 pmol of LebOCH3, 1 pmol of Leb-MCO, and 0.2 pmol of Leb-TMR. Leb-TMR' 1287.4
1273.3
600
800
1000
1
1200
1400
RESULTS AND DISCUSSION It has been shown that underivatized oligosaccharides can be ionized by MALDIF4 However, the limit of detection for underivatized carbohydrates is 100 fmol (for reference standards in clean samples), which is somewhat higher than that obtained for peptides of similar mass.25 For direct analysis of reaction products in serum, the detection sensitivity of MALDI needs to be improved. It has been found that peracetylation of neutral oligosaccharides enhances the detection sensitivity lGfold25 and permethylation enhances the sensitivity of glycolipids 1Wfold.26 After experimenting with different labeling groups, we find that TMR derivatives of oligosaccharides are -100fold more sensitive than underivatized oligosaccharides. Figure 1 shows the MALDI mass spectrum of a mixture of three derivatives of a model Leb blood group active tetrasaccharide: 10 pmol of Leb-OCH3(a simple methyl glycoside derivative), 1pmol of Leb-MCO (a more lipophilic derivative with an 8-methoxycarbonyloctyl aglycon), and 0.2 pmol of Leb-TMR (with the TMR label attached to the MCO group). The spectra shown in Figure 1 are obtained by using 2,5DHB as the matrix. 2,5DHB has been reported to be a superior matrix among several substituted benzoic acids, cinnamic acids, and coumarins, including CHCCA and 3-amino-4-hydroxybenoic acid.25 The neutral sugars in the spectrum show two peaks due to cationization with both sodium and potassium. The peak observed for Leb-TMR
arises from protonation of the TMR zwitterion, although at this time it is unknown whether protonation occurs in the gas phase or if the TMR ion is desorbed preformed (see Chart 1). There is also a predominant peak (at M - 141, possibly due to the loss of CH2 from TMRZ7The signal/background ratios for Leb-OCH3, Leb-MCO,and Leb-TMRare21,82, and 42, respectively. The mass spectra obtained from more concentrated or more diluted solutions of the mixture used in Figure 1 display similar relative detection responses. Leb-TMRis generally about 2-%fold more sensitive than Leb-MCOand lO-fold more sensitive than Leb-OCH3 with 2,5DHB. We find that the cinnamic acid derivative 4-HCCA can also provide good MALDI spectra for the TMR-labeled oligosaccharides. Figure 2 shows the MALDI spectrum of 0.2 pmol of LebTMR acquired using CHCCA as the matrix. The peak at (M 14) intensifies, and additional fragment peaks are observed at m/z 1142.5 (loss of fucose) and 981.5 (loss of fucose and galactose). This finding is not surprising in light of the fact that others have noted fragmentation increases with 4HCCA matrix in reflectron TOF systems.28 Fragmentation can take place during the time-
(24) Stahl, B.; Steup, M.; b a s , M.; Hdlenkamp, F.Anal. Chem. 1991.63,14631466. (25) Harvey, D. J. Rapid Commut~.Mass Spectrom. 1993,7,614-619. (26) Juhasz, P.;Costello, C. E. /. Am. SOC.Mass Spectrom. 1992,3,785-796.
(27) ?he study of several rhodamine derivatives using laser desorption ionization mass spectrometry reveals a predominant peak for the loss of a single alkyl from the alkylamine on tetraalkylrhodamines. Whittal, R M.; Li, L, unpublished results.
MI2 Figure 2. MALDI mass spectrum of 0.2 pmol of Leb-TMR obtained using 4-HCCA as the matrix.
Analytical Chemistry, Vol. 67, No. 19, October 1, 1995
3511
lag period (280 ns) in our instrument. Figure 2 also shows that the signal/background ratio improves to 60, a 50%improvement over that with 2,5DHB. On the other hand, for Leb-OCH3and Leb-MCO,the signal/background ratio drops below 3 if the same quantity of neutral oligosaccharide is loaded, as in Figure 1. Several other TMR-labeled oligosaccharides have been examined, and similar results are obtained. In general, with a sample loading in the range of 100-200 fmol or a concentration of 100-200 fmol pL-', a well-defined molecular ion signal (analyte signal/ background signal > 50) can be obtained for TMR-labeled species using 4HCCA. In order to analyze these labeled species in a buffered serum sample, sample/matrix preparation becomes very important. The MALDI technique is known to be able to tolerate salts and buffers to some extent. This is particularly true for peptide and protein analysi~.~3 The dried droplet method for preparing matrix crystals followed by on-probe sample washing is a simple technique, but it is often not very effective in removing a high concentration of contaminants. In the case of buffered serum, the dried droplet method using either 2,5DHB or 4HCCA fails to form crystals, and a spectrum of only background noise is obtained (not shown). There are several other refined methods reported to wash away potentially interfering contaminants on the probe. In particular, the fast evaporation method of Vorm et and the pressed crystal sample cleanup method developed by Xiang and Beavis30 are effective for removing salts and glycerol from peptide and protein samples. However, using these methods, positive results are not obtained for analyzing oligosaccharidesin buffered serum. For all of the matrices tested, the underlying matrix layer completely redissolves upon deposition of sample and does not recrystallize. We have developed a two-step sample preparation method, which is a modification of the method described by Vorm et al.,29 to analyze labeled oligosaccharides in serum. In preparing the matrix layer on the probe, a high surface density of matrix is necessary to prevent the complete redissolution of the matrix layer when serum is added. Deposition of matrix (Le., CHCCA) from a 20 mg mL-' 70% acetonitrile solution provides small matrix crystals with a high surface density (17 nmol/mm2). To prepare the analyte, an aliquot of serum is diluted with an equal amount of 50% ethanol/water, which precipitates some of the serum proteins. This is an effective first step in the sample analysis procedure, since proteins cannot be readily washed away from the matrix/analyte mixture on the sample probe. The supernatant is then deposited directly on top of the matrix layer. The matrix layer partially redissolves and then recrystallizes, presumably entrapping the analyte. On-probe sample washing then follows. It should be noted that matrices such as 2,5DHB that have a high solubility in aqueous solution are unsuccessful with this approach. For the analysis of the labeled oligosaccharides from serum, this method is found to be facile and effective for removing salts and other contaminants in serum. As an example, Figure 3 shows the MALDI spectrum of a mixture of 30 pmol of Leb-OCH3,10 pmol of Leb-MCO, and 0.5 pmol Leb-TMR spiked into human serum with buffers and other reagents added. Of the three sugars loaded, only Leb-TMR(m/z 1287.4) is detected from this buffered (28) Karas, M.; Bahr, U.; Ehring, H.; Stmpat, IC;Hillenkamp, F. Proceedings of the 42ndASMS Conference on Mass Specfromety and Allied Topics, Chicago, Illinois, May 29-June 3, 1994; p 7. (29) Vorm, 0.; Roepstorff, P.; Matthias, M. Anal. Chem. 1994,66,3281-3287. (30) Xiang, F.;Beavis, R C. Rapid Commun. Mass Spectrom. 1994,8,199-204.
3512 Analytical Chemistry, Vol. 67, No. 79, October 1, 7995
84.9
L~~-TMR'
1287 4 2447.3
o o y 0
, , ,
500
1000
1500
2000
2500
3000
MI2 Figure 3. MALDI mass spectrum of a mixture of 30 pmol of LebOCH3, 10 pmol of Leb-MCO, and 0.5 pmol of Leb-TMR spiked into human serum with buffers and other reagents added. 4-HCCA is used as the matrix with the new samplelmatrix preparation method. LebTMR is the only oligosaccharide derivative observed. The inset graph is an expansion of the peak at m/z 1287.4.
serum sample with a signaVbackground ratio of 11. Therefore, the signal/background ratio for serum samples decreases by a factor of 14 over the nonserum samples. Peaks are observed at m/z 384.9,501.6, and 762.0, along with a significantly weaker peak at 2447.3. These peaks are from unidentified serum components. Other TMR-labeled di-, tri-, and tetrasaccharides have also been studied, and similar findings are obtained. Note that the TMR derivatives are very soluble in water, whereas the 4HCCA matrix is not. The TMR derivatives may preferentially either form cocrystals with the matrix or adsorb onto the matrix crystals. During the sample washing step, salts, buffers, and other potential contaminants are effectively washed away. In the example of Figure 3, Leb-OCH3and Leb-MCOmay also be washed away. This illustratesthat by properly designing the label group in conjunction with an optimal sample/matrix preparation protocol, selective analyte retention and ionization can be achieved. High mass measurement accuracy is essential for unambiguous identification of chemical species on the basis of molecular weight information. In MALDI, the presence of impurities in a sample can potentially degrade the mass resolution and mass accuracy. Thus, an ideal sample preparation method must be able to remove interfering contaminants that would reduce mass accuracy. The mass measurement accuracy for the detection of oligosaccharides was investigated using our linear TOFMS. Figure 4A shows the molecular ion region of the MALDI mass spectrum of the blood group B active TMR-labeled trisaccharide (B-TMR) obtained with a pure reference sample using the 1 gigasample/s digitizer (LeCroy 9350M oscilloscope). The total sample loaded is 200 fmol or 1.0 pL of a sample solution with a concentration of 2 x M. As Figure 4A illustrates,the isotope peaks are well resolved. With a signal/background ratio of 34 for the major isotope peak, the mass resolution is 2600 fwhm. The theoretical exact mass of ETMR is 1099.50 Da. The measured mass of the monoisotopic peak utilizing a two-point external calibration is 1099.45 Da; thus, the error in mass measurement is 46 ppm. For comparison, the mass spectrum of ETMR obtained from buffered serum, using the sample preparation method described above, is shown in Figure 4B. To maintain a comparable signal/background ratio, B-TMR is spiked into
B-TMR' 1099.45
disac-TMR' 936.1
RS = 2600 FWHM
Without UDP-Gal
1.07
4
s
o.4{
/I 1080
1090
1100
1110
1120
800
960.0
1
900
1000
MIZ
p
5 4 s
A
"i 0.5
B,
B-TMR'
1099,55
1100
1200
MI2 disac-TMA' 938.2
Rs = 2100 FWHM
h
0 4
. 203 v) C
5
0.5
g.-
04:
With UDP-Gal
E
E
03-
Q
(1) c
5
1039.4
-c
-
02
.o,
0 1
2
v)
C
02
E0,
8
BTMR' 1 100.3
960.2 01
1039.5
00
MIZ Figure 4. MALDI mass spectra of 6-TMR obtained (A) with a pure sample and (6) from buffered serum spiked with B-TMR. 4-HCCA is used as the matrix with the new samplehatrix preparation method. Spectra were collected with a 1 ns/point digitizer. The theoretical exact mass of 6-TMR is 1099.50 Da.
Figure 5. MALDI for monitoring the aGal(1-3) transferase enzyme reaction in vitro. MALDI mass spectra of the serum samples after a 90-min incubation with no UDP-Gal donor added (A) and with UDPGal donor added (6). 4-HCCA is used as the matrix with the new samplehatrix preparation method.
human serum to a concentration of 6 x lo-' M. One microliter of the sample is loaded to the probe, followed by washing. The MALDI spectrum shown in Figure 4B has a mass resolution of 2100 fwhm. The observed mass of the monoisotopic peak is 1099.55 Da, giving an error in mass measurement of 52 ppm using a two-point external calibration. More generally, repeat measurements (n = 8) show a mass accuracy of 50 f 20 ppm. The calibrants used for Figure 4 are disac-TMR and Leb-TMRdissolved in water at a concentration of 1 x M. This example shows that there is a degradation in mass resolution between the pure reference sample and the buffered serum sample, but not to the point where mass accuracy suffers. Direct monitoring of the in vitro aGal(1-3)transferase (blood group B galactosyl transferase) enzyme reaction is accomplished using this MALDI analysis technique. This enzyme uses UDPGal as the donor and aFuc(l-2)/3Gal-TMR (disac-TMR) as the acceptor to produce ETMR A known amount of disac-TMR (5 nmol in 57.6 p L of serum solution), along with sodium cacodylate buffer (45 mM, pH 7.1) and manganese chloride (5 mM), is added to the serum. The mixture is incubated for 90 min. Figure 5A shows the MALDI spectrum when no UDP-Gal donor is added to the incubation. As expected, product ion is not present in this spectrum. The peaks at m/z 938.1 and 960.0 are for the molecular ions of the starting material disac-TMR due to protonation and sodium cationization, respectively. The peak at m/z 1039.4 is found only in the incubated samples. The origin of this peak is unknown. Figure 5B shows the spectrum taken when 5 m o l of
UDP-Gal donor is added to the starting mixture and incubated for 90 min. In this case, a new peak appears at m/z 1100.3. ETMR is the expected reaction product from the transferase enzyme reaction with a theoretical average mass of 1100.2 Da. For the B-TMR peak, the signal/background ratio is 3, and the intraspectral S/N ratio is 15. This example illustrates that the MALDI technique can be used for monitoring an enzymatic reaction directly from crude serum without the use of traditional timeconsuming separation processes. While this enzymatic reaction can be studied by other tracer techniques, the mass spectrometric technique shown here provides chemical idenscation. In monitoring an enzymatic reaction in a biological sample,there are concerns that the reaction conditions used, or the presence of other unanticipated enzymes, may change the course of the reaction of interest. For example, in the case of the incubation analyzed in Figure 5, fucosidasecatalyzed hydrolysis of the starting material may take place to form free fucose and the TMR-labeled monosaccharideBGal-TMR Radiochemical assay methods would not detect this hydrolysis. HPLC would also fail to identify the products unless mass spectrometry is used for detection. Under the reaction conditions used for Figure 5, there are no other detectable peaks that can be assigned to possible reaction products. The example shown in Figure 5 also illustrates the importance of running a parallel control experiment in identifying the source of the unknown peak at m/z 1039.4. One limitation of the MALDI technique with the current setup is the lack of extensive fragment ions for structural
MI2
Analytical Chemistty, Vol. 67, No. 19, October 1, 7995
3513
1.5 v)
C
0 1.0 C
0.0
0.5 1 .o 1.5 Concentration B-TMR (pmol PL.')
2.0
Figure 6. Calibration curve for quantitation of B-TMR product. The peak intensity of B-TMR is measured relative to Leb-TMR.The error bars represent f2 SD for five samples at each concentration.
analysis. In this regard, MALDI combined with MS/MS can be very powerful in analyzing unexpected peaks. This type of instrument is now commercially available. An ideal mass tracer method should be able to provide quantitative information rapidly. Quantitative information, with regard to the reaction products, can be provided with MALDI if an internal standard is ~sed.4J'~j~~ To quanm the yield of B-TMR formed in Figure 5, a calibration curve was generated using the relative signal intensity of B-TMR with respect to the signal intensity of Leb-TMR The disac-TMR ion intensity was not used because the concentration of disac-TMR changes during incubation. To develop the calibration curve, shown in Figure 6, B-TMR and Leb-TMRare spiked to buffered human serum and analyzed. The concentrationof B-TMR was varied between 0.22 and 2.1 pmol pL-l, while the concentration of Leb-mRwas fixed at 2.1 pmol pL-'. The relative intensity of the molecular ion peak of B-TMR to Leb-TMRis plotted as a function of B-TMR concentration. A linear response is obtained (r = 0.99). The error bars represent f 2 SD for five samples at each concentration. For the sample analysis, LebTMR is spiked to the sample, followed by immediate MALDI analysis. From the calibration curve, the concentration of B-TMR formed after the 90 min incubation of disac-TMR is 0.50 f 0.09 pmol pL-', or a total product yield of 29 f 5 pmol. Also, from the calibration curve, the concentration detection l i t
of B-TMR in human serum can be estimated as 0.11 pmol pL-' (detection limit defined as 3 SD of the background signal level/ sensitivity). It should be noted that the ability to perform analyte quantitation directly from a crude sample as illustrated in this example is very signiicant. This method averts the problems of possible sample loss or quantitative variations often associated with conventional sample cleanup, extraction, and chromatographic procedures. In addition, the calibration curve can be constructed in a much shorter period of time. We are currently in the process of performing a detailed investigation on the quantitative aspect of this mass tracer method. In particular,the linear dynamic range and the rational selection of internal standards in the calibration curve method will be studied. The possibility of using standard addition for quantitation will also be investigated. In addition, a correlation study between the MALDI work and other quantitative methods will be carried out in establishing the validity of these methods. The MALDI approach for the study of enzyme kinetics as well as for the determination of enzyme concentration in crude samples will also be investigated. In summary, we have developed a tracer method based on the MALDI technique for the assay of glycosyltransferases in human serum. We anticipate that this rapid, sensitive, and molecular-specific detection system will be very useful for reaction monitoring and enzyme kinetic studies of oligosaccharides from crude samples including crude cell tissue and organ extracts. This work also demonstrates the importance of the molecular structure and other physical and chemical properties of the analyte in sample preparation or crystal formation in MALDI. We plan to investigate this phenomenon further in the future in order to design better mass tracers and extend this work to other biomolecules. ACKNOWLEWMENT We thank Mr. Paul Diedrich and Mr. Richard Beever for preparing the ThB conjugates.Leb-OCH3and the MCO glycosides were generous gifts from Prof. R U. Lemieux. This work was supported in part by the Natural Sciences and Engineering Research Council of Canada (NSERC). RM.W. thanks NSERC for a postgraduate scholarship. Received for review March 28, 1995. Accepted July 6,
1995.B AC950307U
(31) Nelson, R W.; McLean, M. A; Hutchens, T.W. Anal. Chem. 1994, 66, 1408-1415.
3514 Analytical Chemistry, Vol. 67,No. 19, October 1, 1995
@
Abstract published in Advance ACS Abstracts, August 15, 1995.
